Read the Report.
Appendix A: Additional Notes on State Cottage Food Laws
State cottage food laws are full of idiosyncrasies that cannot be captured by the broad categories displayed in Tables 1 and 2. See below for additional information about the legal factors analyzed in those tables. Note that there may be additional intricacies to a state’s cottage food laws that are not captured in the report and were not considered in the analysis:
Connecticut: Connecticut passed a new cottage food law in 2015, which is reflected in Tables 1 and 2. However, as this report went to print in December 2017, the state had not yet brought the new law into force, so producers were not able to sell their cottage foods under the new regime. Also as this report went to print, the new law was not yet reflected on Forrager.com, so this analysis relies on the text of the law 1 rather than on Forrager.
Delaware: Farmers in Delaware can get a separate “on-farm home processing” license that allows annual sales of up to $40,000. 2
District of Columbia: The District of Columbia passed a new cottage food law in 2013, which is reflected in Tables 1 and 2. 3 However, as this report went to print in December 2017, the Department of Health had not yet created the cottage food registry necessary to allow producers to begin selling their cottage foods legally.
Illinois: Illinois also has a “home kitchen operations” law, which is for bakers and does not require registration. However, it is not available everywhere in the state because counties must specifically adopt it and most have not yet done so. 4 For this reason, the home kitchen operations law is not analyzed in Tables 1 and 2.
Indiana: Producers in Indiana can take orders over the internet, but they must deliver those orders to a farmers’ market or roadside stand for payment. 5
Kentucky: Kentucky’s microprocessors scheme allows the sale of pickles, as well as higher-risk canned goods, such as tomatoes, beans and corn. Because the scheme has such a narrow scope, microprocessors were not included in the survey. 6
Louisiana: Louisiana allows custard or cream-filled bakery products to be sold, provided pasteurized milk products are used to make them, but it does not permit the sale of other refrigerated goods. 7
Maine: Maine’s “food sovereignty law,” adopted in August 2017, is not reflected in this analysis. 8
Maryland: In Maryland, cottage foods may be sold at farmers’ markets or events resembling farmers’ markets: “[a] location in a farmer’s market or at a public festival or event where raw agricultural products … are sold.” 9 Cottage foods may not be sold at other events. 10 For this reason, Maryland is treated as a state that limits cottage food sales to farmers’ markets only.
North Dakota: North Dakota’s food freedom law, passed in January 2017, was not yet reflected on Forrager.com as this report went to print in December 2017. For this reason, the analysis relies on the text of the new law 11 rather than on Forrager.
Ohio: Ohio requires that at least 75 percent of a producer’s honeys and syrups come from the person’s own hives or trees, respectively. 12
Oklahoma: Small-scale honey producers (producing less than 500 gallons per year) in Oklahoma can sell their honey directly to consumers under a law separate from the state’s cottage food law. 13
Oregon: Farmers in Oregon can sell their products under a separate “farm direct” law, as long as they grew the primary ingredients used in the products and limit sales of acidified foods to $20,000 per year. 14
Rhode Island: To be allowed to sell cottage foods, Rhode Island farmers must sell more than $2,500 of agricultural products per year. 15
Vermont: Vermont has several different laws for the sale of homemade foods, so this study focuses on the home baker license. Licensure is required for bakers who sell more than $125 worth of product per week, but the foods and sales venues permitted do not change with licensure. The Vermont home bakers included in this survey were licensed, indicating that they sell (or intend to sell) more than $125 worth of product per week. 16
Virginia: While Virginia does not have a cap on cottage food sales overall, it does have a $3,000 annual sales cap on pickles and other acidified vegetables. 17
Appendix B: Study Methods
Survey
Sample
The final survey sample included 775 cottage food producers across 22 states. The sample was constructed by securing a list of all registered cottage food producers from state, county and local governments in the 25 states whose registration schemes allowed me to identify these producers, listed in Table B1. This facilitated the creation of a population of 25,418 registered cottage food producers. This population does not include people in those states who produce cottage foods illegally. It also does not include people who produce cottage foods legally but who were impossible to identify because they limit their business activity such that they are not required to register. States with multi-tiered regulatory schemes that require some producers to register but not others are marked with an asterisk in Table B1.
The sample was constructed as a stratified random sample. The number of participants from each state was proportional to the percentage of registered cottage food producers from that state in the 25-state registered cottage food producer population. After proportional quota frequencies were set for each state, cottage food producers from the respective state lists were called at random until quotas were filled as close to the target as possible.
Data Collection
To draft the survey instrument, I relied largely on questions from other similar surveys and adapted them for the purpose of this survey. A benefit of this approach was that most of the questions in the survey had already been field tested. Prior to data collection, the survey was pre-tested on a small sample of cottage food producers. Results from the pre-test were used to refine questions for the sake of clarity and precision.
WPA Intelligence, a research company based in the District of Columbia, collected survey data between March 13 and April 6, 2017. In 24 of the 25 states, surveys were completed by telephone. In Arizona, however, surveys were completed online. The state would only release email addresses, not phone numbers or home addresses. The different survey mode in Arizona was controlled for in regression analysis. The full dataset can be found on this website.
Survey Weights
To ensure geographic representativeness of the cottage food producer population and appropriately account for different response rates by producers in different states, a post-survey weighting adjustment was used. The population targets were based on producer counts that were compiled from the 25-state population. Weights were calculated using iterative proportional fitting, which uses a maximum-likelihood algorithm to find the minimum adjustment necessary to make the individual responses match the population distribution of the states.
Table B1: Producers by State
| State | Producers in Population | Percent of Population | Producers in Survey Sample | Weighted Completes |
|---|---|---|---|---|
| Arizona | 5,671 | 22.3% | 103 | 130 |
| California | 2,811 | 11.1% | 90 | 91 |
| Delaware | 6 | 0.0% | 0 | 0 |
| Georgia | 250 | 1.0% | 7 | 7 |
| Illinois* | 362 | 1.4% | 10 | 11 |
| Iowa* | 316 | 1.2% | 25 | 19 |
| Kentucky | 759 | 3.0% | 38 | 27 |
| Maine | 1,285 | 5.1% | 67 | 54 |
| Massachusetts | 661 | 2.6% | 21 | 20 |
| Minnesota | 423 | 1.7% | 32 | 23 |
| Montana | 73 | 0.3% | 0 | 0 |
| Nevada | 161 | 0.6% | 4 | 4 |
| New Hampshire* | 122 | 0.5% | 5 | 4 |
| New York | 3,147 | 12.4% | 92 | 96 |
| North Carolina | 4,186 | 16.5% | 68 | 101 |
| Ohio* | 835 | 3.3% | 33 | 29 |
| Oregon* | 781 | 3.1% | 37 | 32 |
| Pennsylvania | 1,731 | 6.8% | 75 | 62 |
| Rhode Island | 16 | 0.1% | 0 | 0 |
| Tennessee* | 135 | 0.5% | 4 | 4 |
| Utah | 288 | 1.1% | 16 | 15 |
| Vermont* | 167 | 0.7% | 8 | 7 |
| Virginia* | 1,109 | 4.4% | 33 | 34 |
| Washington | 95 | 0.4% | 6 | 4 |
| West Virginia | 28 | 0.1% | 1 | 1 |
| Total | 25,418 | 100% | 775 | 775 |
* Indicates states with multi-tiered regulatory schemes that require some producers to register but not others.
Variable Transformation and Recoding
Several questions were recorded by the surveyors as verbatim text values and had to be recoded into numeric values. Recoding details are contained in Table B2.
Variables marked with an asterisk in Table B2 had high rates of missing values, which posed a problem for the regression analyses. To overcome this problem, I used multiple imputation to impute missing values for use in regression analysis. Descriptive statistics are reported in their original, non-imputed form.
Table B2: Recoding of Numeric Variables
| Variable | Question | Standardized Response | Example of Recoding |
|---|---|---|---|
| Q05 | How long have you worked selling foods you made in your home? | Number of months selling cottage foods | E.g., “seven years and 3 months” became 87. |
| Q12 | How long did it take you to get all the necessary approvals from the government before you could begin selling your homemade foods? | Number of days it took to obtain necessary approvals to operate business | Days obtained by multiplying weeks by 7, months by 30, years by 365. E.g., “6 months” became 180. |
| Q17 | Since beginning your homemade food business, how many times has your home been inspected by the government? | Number of times home inspected | E.g., “four times” became 4. In cases where respondents indicated monthly, annual, etc. inspections, Q05 was used to determine how long they had been in business. The number of inspections was deduced from there. |
| Q28* | During an average work week, how much time do you spend on your homemade food business? | Number of hours spent on the business during an average week | “Days” were treated as 8 hours. “Seasonal” was generally treated as 3 months, so hours indicated were divided by 4. In cases where a range was provided, such as “12–15 hours,” the average was taken. |
| Q29* | How much of that time is spent interacting with customers? | Number of hours spent interacting with customers during an average week | “Days” were treated as 8 hours. “Seasonal” was generally treated as 3 months, so hours indicated were divided by 4. In cases where a range was provided, such as “12–15 hours,” the average was taken. |
| Q30* | How much of that time is spent organizing your homemade food business? | Number of hours spent organizing the business during an average week | “Days” were treated as 8 hours. “Seasonal” was generally treated as 3 months, so hours indicated were divided by 4. In cases where a range was provided, such as “12–15 hours,” the average was taken. |
| Q31* | How many people do you employ full time, not including yourself? | Number of people employ full time | Strictly transferring string values to numeric values. E.g., “two” became 2. |
| Q32* | How many people do you employ part time? | Number of people employ part time | Strictly transferring string values to numeric values. E.g., “two” became 2. |
| Q33* | In 2016: How many dollars did your homemade food business generate in profit, after expenses? | Number of dollars generated in profit last year | Strictly transferring string values to numeric values. E.g., “one thousand dollars” became 1000. |
| Q34* | How many dollars did your homemade food business generate in annual sales, before you deduct expenses? | Number of dollars generated in annual sales last year | Strictly transferring string values to numeric values. E.g., “one thousand dollars” became 1000. |
| Q35* | How much did you pay in sales tax to the city, county or other governments? | Number of dollars paid in sales tax last year | Strictly transferring string values to numeric values. E.g., “one thousand dollars” became 1000. |
| Q36* | How much did you pay for permits, inspections or other fees specifically required to be a homemade food business? | Number of dollars paid for permits, inspections or other fees last year | Strictly transferring string values to numeric values. E.g., “one thousand dollars” became 1000. |
| Q38* | What was the total amount of capital used to start your business? | Number of dollars capital used to start business | Strictly transferring string values to numeric values. E.g., “one thousand dollars” became 1000. |
| Q39 | How long do you plan to continue selling homemade food? | Number of years plan to continue selling | Transferring string values into numeric values. E.g., “ten years” became 10. In cases where respondents said something like “until I retire,” the time between the respondent’s current age and age 65 (average retirement age) was calculated; in cases where the respondent indicated an indefinite time period, the time between the respondent’s current age and age 87 (average life expectancy) was calculated. |
| Q42* | How much was your personal income in 2016? | Number of dollars of personal income last year | Strictly transferring string values to numeric values. E.g., “one thousand dollars” became 1000. |
| Q43* | How much was your household income in 2016? | Number of dollars of household income last year | Strictly transferring string values to numeric values. E.g., “one thousand dollars” became 1000. |
Some variables had skewed distribution. To normalize the distribution for use in regression analysis, I transformed the variables as described in Table B3.
Table B3: Variable Transformations
| Variable | Variable Meaning | Transformation |
|---|---|---|
| Q12 | Number of days it took to obtain necessary approvals to operate business | Large outliers dropped, square root |
| Q29 | Hours spent with customers each week | Natural log+0.0001, to avoid transforming zero values into missing |
| Q30 | Hours spent organizing the business each week | Large outliers dropped |
| Q31 | Number of full-time employees | Natural log+0.0001, to avoid transforming zero values into missing |
| Q32 | Number of part-time employees | Natural log+0.0001, to avoid transforming zero values into missing |
| Q33 | 2016 profits | Natural log+1, to avoid transforming zero values into missing |
| Q34 | 2016 sales | Natural log+1, to avoid transforming zero values into missing |
| Q35 | 2016 sales tax | Natural log+1, to avoid transforming zero values into missing |
| Q36 | Amount paid for permits, inspections and other fees in 2016 | Natural log+1, to avoid transforming zero values into missing |
| Q38 | Amount of startup capital used | Natural log+1, to avoid transforming zero values into missing |
| Q42 | 2016 personal income | Square root |
| Q43 | 2016 household income | Square root |
Some additional variables required deductive coding, inductive coding or a combination of the two. For example, in some cases where the surveyors recorded a response as “other,” it was clear from their verbatim description of the response that the response fit within another response option contained in the survey instrument. In such cases, I used deductive coding to place a response within the variable’s coding scheme. However, some of the “other” responses did not fit within the coding scheme. In those cases, I used inductive coding to group like responses together and used those groupings to formulate additional response options. Finally, some questions did not provide response options and were instead simply recorded verbatim. For these variables, I exclusively used inductive coding to group like responses together and to formulate a coding scheme for use in regression analysis. Explanations of these coding decisions are contained in table B4.
Table B4: Deductive and Inductive Coding
| Variable | Question | Coding |
|---|---|---|
| Q09 | What motivated you to start your cottage food business? | Deductively coded “other” responses into existing response options. Inductively coded those responses that did not fit within the scheme to create the following additional response options: a) I identified a gap in the market and wanted to fill it; b) I identified a good business opportunity; c) Friends and family encouraged me to start selling my foods; d) I wanted to generate additional income during retirement; e) I wanted to use the produce that I was already growing; f) I have a talent for making good food. |
| Q14 | Please tell me what foods you would like to sell, but are prohibited by the government from doing so. | Deductively coded “other” responses into existing response options. Responses that did not fit within the existing coding remained coded as “other.” |
| Q23 | What types of food do you produce? | Deductively coded “other” responses into existing response options. Responses that did not fit within the existing coding remained coded as “other.” |
| Q24 | From what venues or locations do you typically sell? | Deductively coded “other” responses into existing response options. Responses that did not fit within the existing coding remained coded as “other.” |
| Q41 | In what ways do you plan to expand your business? | Inductively coded by grouping like responses together until a coding scheme emerged. The codes were: a) Open brick-and-mortar business; b) Increase sales by acquiring more customers; c) Increase production volume by hiring employees, spending more time on business, investing in larger kitchen or new supplies; d) Other. |
Legal Analysis
Table B5 shows how I coded the cottage food laws of each state in the sample. Some of the states have multi-tiered systems in which some cottage food producers are required to register with the government and some are not. Since I was able to survey only those producers required to register with the government, the analysis below captures the state laws that correspond with required registration. To complement this understanding of a state’s legal environment, producers were also asked, among other questions, how many times their home or point of sale had been inspected, how much they paid in fees to the government in order to operate, and whether they had completed required food handlers’ training.
Table B5: Legal Analysis by State
| State | Sales Cap | Are these producers permitted to sell refrigerated foods? | Are cottage food sales limited to farmers only? | Number of venues where cottage foods may be solda |
|---|---|---|---|---|
| Arizona | None | No | No | 7 |
| California | $50,000 | No | No | Permit A: 5 venues; Permit B: 7 venues |
| Georgia | None | No | No | 5 |
| Illinois | $36,000 | No | No | 1 |
| Iowa | $20,000 | Yes | No | 7 |
| Kentucky | None | No | Yes | 3 |
| Maineb | None | No | No | 7 |
| Massachusetts | None | No | No | 7 |
| Minnesota | $18,000 | No | No | 4 |
| Nevada | $35,000 | No | No | 4 |
| New Hampshire | None | No | No | 7 |
| New York | None | No | No | 3 |
| North Carolina | None | No | No | 7 |
| Ohio | None | Yes | No | 7 |
| Oregon | None | Yes | No | 7 |
| Pennsylvania | None | No | No | 7 |
| Tennessee | None | No | No | 7 |
| Utah | None | No | No | 7 |
| Vermont | None | No | No | 5 |
| Virginia | None | Yes | No | 7 |
| Washington | $25,000 | No | No | 4 |
| West Virginia | None | No | No | 2 |
a I categorized venues based on Forrager.com’s categorization: farmers’ markets, roadside stands, community events, home, online, restaurants and retail stores.
b After the analysis for this report was complete, Maine adopted a new law to allow municipalities to regulate local food distribution, free from state regulatory control. That new law is not reflected in this analysis.
Regression Analysis
The purpose of the analysis was threefold: to determine what effect—if any—legal factors have on 1) cottage food businesses’ annual sales, 2) cottage food producers’ household incomes, and 3) producers’ plans to expand their businesses.
To isolate the effect (β) of legal factors on annual sales and household income, I used ordinary least squares (OLS) regression controlling for a wide array of personal and business characteristics. To isolate the effect (β) of legal factors on the likelihood of producers’ planning to expand their businesses, I employed logistic regression while also controlling for a wide array of personal and business characteristics. For a complete list of control variables used in each analysis, see Table B6.
The primary independent variable in these three analyses was a measure of a state’s sales cap. The measures of a state’s sales cap took three different forms: 1) the dollar amount of the cap, 2) a binary variable that equals 1 if a state has a cap and 0 otherwise, and 3) the sales cap disaggregated into three categories based on the distribution of the sales cap dollar value. Since these three measures did not make a significant difference to the regression results, final results are based only on the dollar amount of the cap as the independent variable.
Regression equations included state probability weights, and standard errors were clustered by state.
The general model for all three analyses was:
Y = β0 + β1(sales_cap) + β2(refrigerated) + β3(venues) + β4(training) + β5(approval) + β6(prohibited_foods) + β7(prohibited_venues) + β8(inspections) + β9(fees) + Θ + Ω + ε
Where:
Model 1: Y = the natural log of a business’s 2016 annual sales (OLS regression)
Model 2: Y = the square root of a producer’s 2016 household income (OLS regression)
Model 3: Y = 1 if a producer plans to expand their business in the near future, 0 otherwise (logistic regression)
In all three models:
sales_cap = the dollar amount of sales cap in state where business operates
refrigerated = 1 if state allows sale of homemade foods requiring refrigeration, 0 otherwise
venues = number of venues (out of seven categories) where state allows cottage foods to be sold
training = 1 if producer was required to undergo training to operate business, 0 otherwise
approval = number of days it took to get government approvals before business could begin
prohibited foods = 1 if there are foods producer wants to sell but is prohibited by government from doing so, 0 otherwise
prohibited venues = 1 if there are venues where producer wants to sell, but is prohibited by government from doing so, 0 otherwise
inspections = number of times home has been inspected by government
fees = natural log of dollar amount paid for permits, inspections or other fees specifically required to sell cottage foods
Θ = business characteristics (see Table B6)
Ω = personal characteristics (see Table B6)
ε = error term
Model 3 also used an interaction term, prohibited foods*rural = 1 if the producer lives in a rural area and there are foods the producer wants to sell but is prohibited by government from doing so, 0 otherwise.
Table B6: Control Variables included in regression models
| Control Variable | Definition | Model 1 | Model 2 | Model 3 |
|---|---|---|---|---|
| Business characteristics | ||||
| main_occupation | =1 if business is a main occupation, compared to a supplemental occupation, 0 otherwise | x | x | x |
| hobby | =1 if business is a hobby, compared to a supplemental occupation, 0 otherwise | x | x | x |
| hours | Number of hours spent on business per week | x | x | x |
| full_employees | Natural log of number of full-time employees | x | x | x |
| part_employees | Natural log of number of part-time employees | x | x | x |
| annual_sales | Natural log of dollar amount of 2016 annual sales | x | x | |
| capital | Natural log of dollar amount of capital used to start the business | x | ||
| continue_selling | Number of years respondent plans to continue selling cottage foods | x | ||
| importance | Producer’s ranking of how important the business is to the financial well-being of their household, 1–6, 6 being most important | x | x | x |
| baked | =1 if respondent sells baked goods, 0 otherwise | x | x | x |
| confectionary | =1 if respondent sells confectionary, 0 otherwise | x | x | x |
| condiments | =1 if respondent sells condiments, 0 otherwise | x | x | x |
| dry_goods | =1 if respondent sells dry goods, 0 otherwise | x | x | x |
| pastries | =1 if respondent sells pastries, 0 otherwise | x | x | x |
| preserves | =1 if respondent sells preserves, 0 otherwise | x | x | x |
| snacks | =1 if respondent sells snacks, 0 otherwise | x | x | x |
| sell_refrigerated | =1 if respondent sells refrigerated goods, 0 otherwise | x | x | x |
| farmers_markets | =1 if respondent sells at farmers’ markets, 0 otherwise | x | x | x |
| roadside_stands | =1 if respondent sells at roadside stands, 0 otherwise | x | x | x |
| community_events | =1 if respondent sells at community events, 0 otherwise | x | x | x |
| home | =1 if respondent sells from home, 0 otherwise | x | x | x |
| restaurants | =1 if respondent sells at restaurants, 0 otherwise | x | x | x |
| retail_stores | =1 if respondent sells at retail stores, 0 otherwise | x | x | x |
| online_phone | =1 if respondent sells online or by phone, 0 otherwise | x | x | x |
| Personal characteristics | ||||
| personal_income | Square root of dollar amount of 2016 personal income | x | x | |
| household_income | Square root of dollar amount of 2016 household income | x | ||
| race | =1 if respondent is white, 0 otherwise | x | x | x |
| married | =1 if respondent is married, 0 otherwise | x | x | x |
| education | Respondent’s level of education, ranked 1–5, 5 being highest | x | x | x |
| children | =1 if there are children under the age of 18 in the respondent’s household, 0 otherwise | x | x | x |
| rural | =1 if the respondent lives in a rural area, compared to suburban, 0 otherwise | x | x | x |
| urban | =1 if the respondent lives in an urban area, compared to suburban, 0 otherwise | x | x | x |
| gender | =1 if female, 0 if male | x | x | x |
| arizona | =1 if respondent lives in Arizona, 0 otherwise (this controls for the different survey mode employed in Arizona) | x | x | x |
Appendix C: Regression Results
Table C1: Model 1
| Coefficient | Robust Clustered S.E. | ρ | |
|---|---|---|---|
| sales_cap | 0.000 | 0.000 | 0.867 |
| refrigerated | 0.290 | 0.391 | 0.473 |
| venues | -0.035 | 0.088 | 0.696 |
| training | -0.157 | 0.448 | 0.731 |
| approval | 0.005 | 0.040 | 0.911 |
| prohibited_foods | -0.304 | 0.303 | 0.332 |
| prohibited_venues | 0.066 | 0.336 | 0.847 |
| inspections | 0.016 | 0.022 | 0.493 |
| fees | 0.147 | 0.074 | 0.070 |
| main_occupation | 1.000 | 0.545 | 0.086 |
| hobby | -0.782 | 0.352 | 0.042 |
| hours | 0.004 | 0.011 | 0.723 |
| f_employees | -0.028 | 0.061 | 0.656 |
| p_employees | 0.022 | 0.045 | 0.628 |
| importance | 0.071 | 0.114 | 0.545 |
| baked | 0.086 | 0.355 | 0.812 |
| confectionary | -0.182 | 0.470 | 0.704 |
| condiments | 0.613 | 0.508 | 0.247 |
| dry_goods | -0.138 | 0.643 | 0.833 |
| pastries | 0.606 | 0.602 | 0.332 |
| preserves | -0.918 | 1.013 | 0.381 |
| snacks | 0.226 | 1.023 | 0.828 |
| sell_refrigerated | 0.504 | 1.027 | 0.632 |
| farmers_markets | 0.109 | 0.347 | 0.761 |
| roadside_stands | 0.315 | 0.620 | 0.620 |
| community_events | -0.116 | 0.427 | 0.789 |
| home | 0.200 | 0.337 | 0.563 |
| restaurants | -0.037 | 0.481 | 0.940 |
| retail_stores | 0.586 | 0.403 | 0.184 |
| online_phone | 0.315 | 0.482 | 0.525 |
| personal_income | 0.006 | 0.002 | 0.002 |
| race | 0.041 | 0.498 | 0.935 |
| married | 0.433 | 0.361 | 0.252 |
| education | -0.001 | 0.133 | 0.993 |
| children | 0.047 | 0.298 | 0.877 |
| rural | 0.146 | 0.386 | 0.711 |
| urban | -0.146 | 0.586 | 0.808 |
| gender | 0.228 | 0.471 | 0.635 |
| Arizona | -1.056 | 0.389 | 0.016 |
| intercept | 4.820 | 1.266 | 0.002 |
Table C2: Model 2
| Coefficient | Robust Clustered S.E. | ρ | |
|---|---|---|---|
| sales_cap | 0.000 | 0.000 | 0.908 |
| refrigerated | 28.428 | 18.727 | 0.168 |
| venues | 1.829 | 4.265 | 0.677 |
| training | 8.331 | 13.357 | 0.543 |
| approval | -1.610 | 1.663 | 0.355 |
| prohibited_foods | 0.004 | 9.112 | 1.000 |
| prohibited_venues | 13.924 | 16.223 | 0.410 |
| inspections | -0.442 | 0.894 | 0.634 |
| fees | 5.233 | 3.235 | 0.131 |
| main_occupation | 11.372 | 19.598 | 0.572 |
| hobby | -10.822 | 15.027 | 0.483 |
| hours | -0.597 | 0.389 | 0.153 |
| f_employees | -1.125 | 2.394 | 0.651 |
| p_employees | 0.793 | 1.993 | 0.697 |
| annual_sales | 8.160 | 2.259 | 0.005 |
| importance | -17.680 | 4.413 | 0.003 |
| baked | 23.066 | 14.186 | 0.130 |
| confectionary | -15.042 | 21.052 | 0.490 |
| condiments | 27.768 | 18.219 | 0.160 |
| dry_goods | 5.723 | 21.992 | 0.799 |
| pastries | -15.036 | 25.049 | 0.559 |
| preserves | 11.650 | 35.044 | 0.747 |
| snacks | -17.078 | 33.570 | 0.618 |
| sell_refrigerated | -15.768 | 36.726 | 0.680 |
| farmers_markets | -7.505 | 14.995 | 0.625 |
| roadside_stands | 14.981 | 23.115 | 0.532 |
| community_events | 16.213 | 14.867 | 0.293 |
| home | 9.421 | 12.370 | 0.462 |
| restaurants | 11.775 | 20.160 | 0.567 |
| retail_stores | -1.476 | 17.533 | 0.935 |
| online_phone | 10.383 | 20.406 | 0.621 |
| race | -0.159 | 17.521 | 0.993 |
| married | 27.652 | 13.087 | 0.060 |
| education | 20.715 | 4.899 | 0.002 |
| children | 11.302 | 12.874 | 0.399 |
| rural | 16.131 | 16.480 | 0.358 |
| urban | 7.616 | 24.874 | 0.766 |
| gender | -1.909 | 19.011 | 0.921 |
| Arizona | 20.976 | 21.578 | 0.373 |
| intercept | 41.983 | 50.236 | 0.426 |
Table C3: Model 3
| Coefficient | Robust Clustered S.E. | ρ | |
|---|---|---|---|
| sales_cap | 0.000 | 0.000 | 0.702 |
| refrigerated | -0.235 | 0.349 | 0.501 |
| venues | 0.037 | 0.061 | 0.542 |
| training | 0.139 | 0.429 | 0.745 |
| approval | 0.012 | 0.029 | 0.683 |
| prohibited_foods | 1.300 | 0.294 | 0.000 |
| prohibited*rural | -1.123 | 0.396 | 0.005 |
| prohibited_venues | 0.509 | 0.307 | 0.097 |
| inspections | -0.031 | 0.020 | 0.121 |
| fees | 0.023 | 0.049 | 0.636 |
| main_occupation | 0.073 | 0.385 | 0.850 |
| hobby | 0.678 | 0.225 | 0.003 |
| hours | 0.006 | 0.009 | 0.492 |
| f_employees | 0.066 | 0.041 | 0.105 |
| p_employees | 0.012 | 0.030 | 0.697 |
| annual_sales | -0.067 | 0.045 | 0.134 |
| capital | 0.128 | 0.060 | 0.033 |
| continue_selling | 0.036 | 0.008 | 0.000 |
| importance | 0.249 | 0.099 | 0.012 |
| baked | -0.169 | 0.369 | 0.648 |
| confectionary | -0.421 | 0.365 | 0.249 |
| condiments | 0.572 | 0.364 | 0.116 |
| dry_goods | 0.206 | 0.565 | 0.716 |
| pastries | -0.821 | 0.372 | 0.027 |
| preserves | 1.696 | 0.889 | 0.056 |
| snacks | 0.101 | 0.707 | 0.887 |
| sell_refrigerated | 1.618 | 0.678 | 0.017 |
| farmers_markets | 0.093 | 0.230 | 0.686 |
| roadside_stands | 0.855 | 0.514 | 0.096 |
| community_events | -0.123 | 0.217 | 0.570 |
| home | -0.157 | 0.270 | 0.562 |
| restaurants | 0.252 | 0.392 | 0.521 |
| retail_stores | 0.637 | 0.298 | 0.032 |
| online_phone | 0.582 | 0.448 | 0.194 |
| personal_income | -0.001 | 0.002 | 0.587 |
| household_income | 0.000 | 0.001 | 0.860 |
| race | -1.324 | 0.314 | 0.000 |
| married | 0.109 | 0.364 | 0.765 |
| education | 0.266 | 0.078 | 0.001 |
| children | 0.682 | 0.209 | 0.001 |
| rural | -0.109 | 0.329 | 0.740 |
| urban | 0.874 | 0.515 | 0.090 |
| gender | -0.814 | 0.418 | 0.051 |
| Arizona | 0.348 | 0.468 | 0.457 |
| intercept | -2.088 | 1.070 | 0.051 |
Appendix D: Descriptive Statistics
The following tables provide descriptive statistics for the sample that were not otherwise presented in the main text of the report.
Descriptive Statistics
| Race/Ethnicity | |
|---|---|
| White | 83.8% |
| Hispanic, Mexican, Latino, Spanish | 3.1% |
| African-American | 6.4% |
| Asian | 1.6% |
| Other | 2.6% |
| Refused | 2.5% |
| Highest Level of Education | |
|---|---|
| Less than high school graduate | 3.8% |
| High school graduate | 16.9% |
| Some college/associate’s degree | 37.1% |
| Bachelor’s degree | 25.7% |
| Post-graduate | 15.7% |
| Don’t know/Refused | 0.8% |
| Marital Status | |
|---|---|
| Single, never married | 13.0% |
| Married | 71.4% |
| Separated | 0.9% |
| Divorced | 8.5% |
| Widowed | 4.7% |
| Don’t know/Refused | 1.4% |
| Children in Household Under Age of 18 | |
|---|---|
| Yes | 34.8% |
| No | 54.5% |
| Don’t know/Refused | 10.8% |
| Recognized Disability | |
|---|---|
| Yes | 8.8% |
| No | 89.2% |
| Don’t know/Refused | 2.0% |
| Provide Care for Disabled, Sick, Elderly or Otherwise Incapacitated Person | |
|---|---|
| Yes | 11.0% |
| No | 87.9% |
| Don’t know/Refused | 1.1% |
| Respondent/Spouse Currently Serving in the Military | |
|---|---|
| Yes | 0.6% |
| No | 98.4% |
| Don’t know/Refused | 0.9% |
| Age of Respondents | |
|---|---|
| 18–24 | 1.2% |
| 25–34 | 11.4% |
| 35–44 | 17.1% |
| 45–54 | 23.6% |
| 55–64 | 25.3% |
| 65–74 | 13.4% |
| 75+ | 4.0% |
| Refused | 3.9% |