Part A - Interpreting Model Outputs

You manage a community-based organization with an aim to increase food security in the inner city. You recently hired a consultant to conduct a survey of the factors that drive food insecurity, measured as the number of times a person reports they went without a meal in the past month. The consultant collected all the data and ran several statistical outputs. Unfortunately, budget cuts meant that you could not have the consultant produce a summary report. Rather, what you have are the following outputs from a multiple regression model.

The sample of 2853 was collected from parents in household with a range of 0 to 6 children.

LINEAR MODEL SUMMARY
R-Squared	.390
Adjusted R-squared	.380
Sample Size	2853
ANOVA		Significance
F statistic	85.35	.002
MODEL	Beta	Standardized Beta	T statistic	Significance	Lower Bound	Upper Bound
Intercept	8		5.5	.001	6.5	10.7
Monthly income in dollars	-0.0015	-11.56	1.23	.001	-0.23	-.05
Kilometers to nearest food bank	1.35	0.25	8.2	.005	-0.45	-0.67
Monthly receipt of good food book (0=no, 1=yes)	-4.8	-8.35	12.56	.002	-6.8	-2.4
Number of children in household	2.5	3.8		.001	1.3	3.8
Kilometers to public library	0.25	0.05	6.23	.178	-.01	.60
Hours worked in paid employment	-0.45	-2.3	5.85	.003	-.14	-.06

Question A1 - What method was used to study the factors driving food insecurity? (2 marks)

Question A2 - How much variance in food insecurity is explained by the model? Is this model helpful in explaining food insecurity? (3 marks)

Question A3 - Explain the impact monthly income has on food insecurity? Does that relationship make sense to you? Why or why not? (4 marks)

Question A4 - Explain the impact number of children has on food insecurity? Does that relationship make sense to you? Why or why not? (4 marks)

Questions A5 - Explain the impact distance to public library has on food insecurity? Does that relationship make sense to you? Why or why not? (4 marks)

Question A6 - What factor (variable) has the most impact on food insecurity? What information from the output did you use to determine this? (2 marks)

Question A7 - What is the regression co-efficient for "monthly receipt of good food box"? What impact does receiving the good food box have on food insecurity? (3 marks)

Question A8 - A new male client walks in your door. The client does not recall how many meals they missed in the last month. However, they can give you the following information.

• Monthly income: 200
• Kilometers to closest food bank: 8.5
• Does not receive the monthly good fox box
• Has 4 children in the household
• Lives 0.8 kilometers from the nearest public library
• Does not work any hours in paid employment

What is the predicted value for the number of meals this person would have missed in the month? Show your work. Based on the idea that people eat three meals a day, does this value seem sensible to you? Why or why not? (5 marks)

Question A9 - A female client walks in now with all the same characteristics, but she has 9 children at home. Would you use the model to predict the number of meals she missed? If so, what is the value? If not, why not? (3 marks)

Part B - Critical Engagement of Model Output

The consultant is upset she won't get a chance to write up the results and receive the rest of the contract money. Nevertheless, since she really appreciates the cause of your organization, she has agree to offer you a free consultation call to ask her for additional information.

Question B1 - What additional charts would ask for from the consultant to ensure the model used was appropriate? Why? (2 marks)

Question B2 - What are three additional questions would you ask of the consultant to see if the approach used here was valid? (3 marks)

Part C - Applying Effect Sizes

You are interested in determining if the average number of meals missed is higher or lower between persons in the community who attend your program and those who do not. You collect a sample from both groups. The first sample of people who use your program is 15 people, the second sample of people who do not use your program is 22 people. The average for people who use your program is 12.57 and the average for people who do not use your program is 15.89.

Question C1 - What statistical test are you going to use to ensure that your results are statistically significance and not due to random chance that comes from sampling? (2 marks)

Question C2 - What effect size statistic would you compute if you got statistically significant results? (2 marks)

Question C3 - You do the calculations. You get a statistically significant result showing the averages are different and your effect size statistic equals 0.53. Describe the impact of your program below and why you think that is the impact. Do this in one or two sentences. (3 marks)

Part D - Forecasts

You work in the Ministry of Agriculture and want to help senior leaders understand what future crop production scenarios might look like.

Question D1 - Using your crop production file from the previous labs. Forecast total crop production out to 2045. Use exponential smoothing. Copy and paste the chart you produce below? (3 marks)

Question D3 - You Deputy Minister asks you what the expected total crop production from your model would be for the year 2037. What would you tell him? (2 marks)

Question D2 - What are some concerns you have about using this data to forecast to 2045? (3 marks)