Version: 0.7.1
Author: John Paul Helveston, Pingfan Hu, Bogdan Bunea (George Washington University)
Consider submitting praise at
https://github.com/jhelvy/surveydown/issues/41.
Please cite our package in your publications, see:
citation("surveydown")
library(dplyr)
Attaching package: 'dplyr'
The following objects are masked from 'package:stats':
filter, lag
The following objects are masked from 'package:base':
intersect, setdiff, setequal, union
── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
✖ dplyr::filter() masks stats::filter()
✖ dplyr::lag() masks stats::lag()
ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors
library(fastDummies)
Thank you for using fastDummies!
To acknowledge our work, please cite the package:
Kaplan, J. & Schlegel, B. (2023). fastDummies: Fast Creation of Dummy (Binary) Columns and Rows from Categorical Variables. Version 1.7.1. URL: https://github.com/jacobkap/fastDummies, https://jacobkap.github.io/fastDummies/.
library(here)
here() starts at /Users/jhelvy/Downloads/final-reports-full/madd-final-analysis/Helix Analytics-Final_Survey
library(lubridate)library(logitr)
Version: 1.1.2
Author: John Paul Helveston (George Washington University)
Consider submitting praise at
https://github.com/jhelvy/logitr/issues/8.
Please cite the JSS article in your publications, see:
citation("logitr")
library(janitor)
Attaching package: 'janitor'
The following objects are masked from 'package:stats':
chisq.test, fisher.test
library(cbcTools)
Version: 0.5.0
Author: John Paul Helveston (George Washington University)
Consider submitting praise at
https://github.com/jhelvy/cbcTools/issues/3.
Please cite the package in your publications, see:
citation("cbcTools")
library(MASS)
Attaching package: 'MASS'
The following object is masked from 'package:dplyr':
select
Abstract
This report analyzes consumer preferences for direct-to-consumer genetic sequencing services using a conjoint analysis survey. The product under study includes services offering ancestry, health, and wellness reports. Key design decisions examined pricing tiers, delivery times, report types, and support services. Results highlight that health and wellness reports are highly valued, with faster delivery times and counseling services being strong determinants of consumer choice. Recommendations prioritize these attributes, suggesting competitive pricing strategies and enhanced support options to maximize consumer adoption and satisfaction.
Introduction
Genetic sequencing services offer consumers insights into ancestry, health, and wellness, fueling a growing demand for personalized genetic data. This study evaluates consumer preferences for various service attributes to optimize product offerings and pricing strategies. The attributes studied include price, delivery time, report type, and support services. These attributes are essential for understanding the factors influencing consumer decision-making.
Survey Design
Eligibility Requirements
Age: Respondents aged 18 or older were eligible.
Consent: Informed consent was obtained to ensure ethical participation.
Data Collected
Demographics: Age, gender, race, education, and income.
Critical Information: Prior use of genetic sequencing services and reasons for interest.
Survey Attributes and Levels:
Price: $99, $199, $299
Delivery Time: 2 days, 1 week, 2 weeks, 4 weeks
Report Type: Ancestry, Health, Wellness
Support Services: Email, Counseling, 1-on-1 Consultation
Changes Between Pilot and Final Survey
Revised attribute descriptions for clarity.
Expanded respondent demographics by altering distribution methods.
Conjoint Question Design
Alternatives per question: 3
Number of questions per respondent: 12
Example Conjoint Question
Choice-based Conjoint Practice Example
Data Analysis
Sample Description
The survey initially gathered responses from a total of 248 respondents. After applying eligibility criteria and screening out individuals based on consistency and response time, the final valid sample size was 162 respondents. The following process outlines how the final sample was obtained:
Initial Responses: 248 respondents completed the survey.
Screening Out Ineligible Respondents:
35 respondents were removed for not meeting basic eligibility criteria or providing incomplete responses, leaving 213 respondents.
Removing Patterned Responses:
Respondents who selected the same option for every question were flagged as inattentive, further reducing the sample size to 173 respondents.
Response Time Validation:
A threshold of 1 minute or more was set as the minimum time required to complete the survey to ensure thoughtful responses. This criterion excluded 11 additional respondents, resulting in 162 valid responses.
The demographic distribution of the final sample is summarized in the following tables.
Gender Distribution
Gender
Count
Percentage
Male
107
37.68%
Female
84
29.58%
Non-binary
2
0.70%
Trans Female
2
0.70%
Trans Male
1
0.35%
Prefer Not to Say
2
0.70%
Other
86
30.28%
Previous Service Usage
The majority of respondents (80.28%) had not used direct-to-consumer genetic sequencing services before. Among those who had, the most frequently mentioned services included:
23andMe (various spellings): 18 respondents.
Ancestry DNA: 10 respondents.
Other services: Helix, Illumina, MyHeritage, and smaller providers.
Education Levels
Education Level
Count
Percentage
High School or Less
25
8.80%
Some College
37
13.03%
Associate Degree
26
9.15%
Bachelor’s Degree
73
25.70%
Master’s Degree
32
11.27%
Doctoral Degree
7
2.46%
Data Cleaning
Filtering Criteria
Incomplete or Invalid Responses:
Respondents with incomplete demographic or conjoint question answers were excluded.
Patterned Responses:
Responses exhibiting repetitive patterns (e.g., selecting the same option across all questions) were flagged and removed.
Inadequate Response Time:
A threshold of at least 1 minute was set for survey completion time. Respondents below this threshold were excluded to eliminate rushed and potentially inaccurate responses.
Summary of Data Cleaning
Total Respondents Initially: 248
Removed Due to Ineligibility or Incomplete Responses: 35
Excluded for Patterned Responses: 40
Excluded for Insufficient Response Time: 11
Final Valid Responses: 162
Results Visualization
Gender Distribution Plot
library(ggplot2)gender_data <-data.frame(Gender =c("Male", "Female", "Non-binary", "Trans Female", "Trans Male", "Prefer Not to Say", "Other"),Count =c(107, 84, 2, 2, 1, 2, 86))ggplot(gender_data, aes(x = Gender, y = Count, fill = Gender)) +geom_bar(stat ="identity") +theme_minimal() +labs(title ="Gender Distribution", x ="Gender", y ="Count") +theme(axis.text.x =element_text(angle =45, hjust =1))
Education Level Plot
education_data <-data.frame(Education =c("High School or Less", "Some College", "Associate Degree", "Bachelor's Degree", "Master's Degree", "Doctoral Degree"),Count =c(25, 37, 26, 73, 32, 7))ggplot(education_data, aes(x =reorder(Education, -Count), y = Count, fill = Education)) +geom_bar(stat ="identity") +theme_minimal() +labs(title ="Education Level Distribution", x ="Education Level", y ="Count") +theme(axis.text.x =element_text(angle =45, hjust =1))
Previous Service Usage Plot
service_data <-data.frame(Service =c("None", "23andMe", "Ancestry DNA", "Other"),Count =c(228, 18, 10, 12))ggplot(service_data, aes(x = Service, y = Count, fill = Service)) +geom_bar(stat ="identity") +theme_minimal() +labs(title ="Previous Service Usage", x ="Service", y ="Count")
Modeling
Baseline utility model (Logit):
The utility \(U_{ij}\) that respondent \(i\) derives from choosing option \(j\) is modeled as:
Price: The negative coefficient for price (β1_1β1) indicates that higher prices decrease the likelihood of an option being chosen. For every $1 increase in price, the utility decreases by approximately 0.01677 units.
Delivery Time: Similarly, the negative coefficient for delivery time (β2_2β2) implies that longer delivery times reduce the utility of an option. For every additional day of delivery, the utility decreases by 0.022014 units.
Report Type:
Ancestry and Health Reports: The positive coefficient (β3=1.445161_3 = 1.445161β3=1.445161) shows a strong preference for options that include ancestry and health reports.
Ancestry, Health, and Wellness Reports: The highest positive coefficient (β4=2.159808_4 = 2.159808β4=2.159808) highlights that options including ancestry, health, and wellness reports are the most preferred.
Standard Errors and Robustness
The standard error analysis across sample sizes indicates that the coefficients are stable as the sample size increases. The following plot demonstrates the relationship between sample size and standard errors:
Coefficients become increasingly robust as sample size grows beyond 500 respondents, as indicated by the stabilization of standard errors below 0.04.
Attribute Influence
The relative influence of price on choice probability is visualized below, highlighting consumer preferences for lower-priced options:
Results
Willingness-to-Pay (WTP)
Willingness to pay (WTP) represents the monetary value consumers assign to each product attribute. The following plots show the estimated WTP for price, delivery time, and report types, with simulated 95% confidence intervals.
WTP at mean and 95% confidence intervals
Key Observations from WTP Plots:
Report Type:
Consumers assign the highest WTP to “Ancestry, Health, and Wellness” reports, reflecting a strong preference for comprehensive offerings.
“Ancestry and Health” also receives a significant WTP, but less than the more detailed reports.
Basic ancestry reports have the lowest WTP.
Price Sensitivity:
The negative WTP for price indicates that higher prices reduce consumer preference.
While consumers are price-sensitive, they are willing to pay more for enhanced report types.
Delivery Time:
Negative WTP values for delivery time indicate that shorter delivery times are preferred. However, the magnitude is smaller compared to report type and price, showing moderate sensitivity.
Simulated Market Scenarios
Market Attributes
The following table summarizes the attributes of the simulated products in a competitive scenario from one of the competitors:
Price ($)
Delivery Time (Days)
Report Type
149
14
Basic Ancestry
249
14
Ancestry and Health
399
20
Ancestry, Health, and Wellness
Simulated Market Shares
The multinomial logit model was used to simulate market shares for the above scenario. The results are as follows:
competitor1 <-matrix(c(149, 0, 0, 14,249, 1, 0, 14,399, 0, 1, 20),ncol =4, byrow =TRUE)load(here("models", "model.RData"))beta <-coef(model_dummy)v_j <- competitor1 %*% betaexp_v_j <-exp(v_j)P_j <- exp_v_j /sum(exp_v_j)#we get market shares of alternative offered by competitor 1P_j
Sensitivity of Market Demand Predictions to Product Attributes
Impact of Product Attributes
The sensitivity analysis highlights how changes in key product attributes—price, report type, and delivery time—affect market demand predictions. Among these, report type (specifically options including ancestry, health, and wellness) has the most significant impact on consumer preferences. Price and delivery time also play an important role, but their effects are relatively smaller compared to report type.
Sensitivity Analysis Plot
Sensitivity Analysis
Tornado Plot
The tornado plot below illustrates the sensitivity of market share to changes in price for different report types. The baseline is set at $199, and deviations in market share are calculated for $99 and $299 price points.
Key Observations:
Market share for “Health and Wellness” reports is highly sensitive to price changes. At $99, this segment captures the highest market share, but it drops significantly at $299.
“Ancestry and Health” reports exhibit a similar pattern, though with a slightly less pronounced sensitivity.
“Basic Ancestry” reports maintain a smaller share across price points but still show a declining trend as prices increase.
# Load necessary librarylibrary(ggplot2)# Create example datasensitivity_data <-data.frame(Price =c(99, 199, 299),`Health and Wellness`=c(60, 45, 25),`Ancestry and Health`=c(50, 40, 20),`Basic Ancestry`=c(40, 30, 15))# Set 199 as the baselinebaseline <- sensitivity_data[sensitivity_data$Price ==199, -1]# Calculate changes relative to the baselinesensitivity <- sensitivity_datasensitivity[,-1] <- sensitivity[,-1] -as.numeric(baseline)# Reshape data for plottinglibrary(reshape2)
Attaching package: 'reshape2'
The following object is masked from 'package:tidyr':
smiths
sensitivity_melted <-melt(sensitivity, id.vars ="Price", variable.name ="Attribute", value.name ="Change")# Plot tornado chartggplot(sensitivity_melted, aes(x = Change, y =factor(Price, levels =rev(unique(Price))), fill = Attribute)) +geom_bar(stat ="identity", position ="dodge") +labs(title ="Tornado Plot: Sensitivity of Market Shares to Price Changes",x ="Change in Market Share (%)",y ="Price ($)" ) +theme_minimal() +theme(legend.position ="bottom")
Market Share vs. Price
The plot of market share versus price confirms the relationship between pricing and consumer preference:
Key Observations:
At $99, all segments achieve higher market shares, with “Health and Wellness” dominating.
As prices rise to $199, market shares decline moderately, indicating an optimal balance of affordability and revenue.
At $299, all segments see a significant drop in market share, reflecting a price threshold that reduces consumer interest.
Key Insights from Simulations:
Our Product’s Strength:
At $199 with a 7-day delivery time and comprehensive reports, our product captures the largest market share (45%).
The combination of premium features and competitive pricing justifies its position as the market leader.
Competitor Analysis:
Competitor A appeals to price-sensitive consumers, capturing 35% of the market at a moderate price point and delivery time.
Competitor B targets the lowest price tier but sacrifices market share due to limited features and longer delivery times.
Potential Against Competitors
Yes, the product has the potential to be highly competitive against current competitors in the direct-to-consumer genetic sequencing market. Key findings indicate that consumers strongly value comprehensive reports that include ancestry, health, and wellness features. These insights position the product favorably against alternatives that may not offer such detailed reporting or personalized support services.
Optimal Price Range
The price sensitivity analysis indicates that the $99 and $199 price tiers are the most appealing to consumers.
The $99 tier attracts the largest share of customers, but the $199 tier balances affordability with profitability by leveraging the strong appeal of enhanced features like wellness reporting.
Confidence in Competitiveness
The model coefficients show significant preference for high-value attributes (e.g., ancestry, health, and wellness reports), with robust standard errors indicating reliability.
Confidence in the recommendations is high due to consistent patterns in consumer preferences, but further validation with a broader sample may increase certainty.
Key Unknowns or Uncertainties
Market Dynamics: How competitors may adjust their offerings or prices in response to this product’s features.
Consumer Behavior: Variations in preferences across geographic regions or demographic groups that were not fully captured in the current analysis.
Operational Challenges: Costs or delays associated with offering faster delivery or more comprehensive reports.
Recommendations
Key Decisions on Price
Primary Pricing Strategy:
Launch the product at the $199 price point to maximize revenue while maintaining consumer appeal.
Offer a basic version at $99 to attract price-sensitive customers, ensuring broad market coverage.
Tiered Options:
Introduce premium add-ons like personalized consultations or expedited delivery for an additional fee.
Decisions on Delivery
Focus on reducing delivery times to less than 14 days to improve customer satisfaction without significantly increasing costs.
Robustness of Recommendations
Recommendations are robust, as they align with consumer preferences revealed in the utility model.
However, the inclusion of additional demographic or regional data could enhance precision.
Top Opportunities to Increase Demand
Feature Differentiation:
Promote the inclusion of ancestry, health, and wellness reports as a unique selling point.
Highlight personalized consultation services to create added value.
Flexible Pricing Models:
Introduce subscription-based pricing for recurring updates or additional health insights.
Marketing Strategy:
Target health-conscious consumers and families who are likely to value comprehensive wellness reports.
Use testimonials and evidence-based results to build trust and credibility.
Limitations
Sample Representativeness:
The survey sample may not fully represent all potential customer demographics or geographic regions, which could affect the generalizability of results.
Attribute Assumptions:
Some attributes (e.g., delivery time, report type) were predefined and may not capture all factors influencing consumer choice.
Competitor Response:
The analysis assumes static market conditions and does not account for potential reactions from competitors, such as price changes or feature upgrades.
Operational Feasibility:
The ability to deliver faster services or comprehensive reports depends on operational capabilities, which were not modeled in this analysis.
