Appendix 9_Sample Size Calculation Table

APPENDIX 9

Sample Size Calculations – Number of cases needed

Change in Exposure (Behavior) in CASE Group								Frequency of Exposure (Behavior) in CONTROL Group	Change in Exposure (Behavior) in CASE Group
-11%	-10%	-9%	-8%	-7%	-6%	-5%	-4%		4%	5%	6%	7%	8%	9%	10%	11%
						474	770	10%		725	525
					492	725		15%			675	510
					647	945		20%			804	603
				568	780			25%				680	529
			495	651				30%				741	574
			548	718				35%				786	607
			589	770				40%				815	628
			617	805				45%				827	637
			633	824				50%				824	633
			637	827				55%				805	617
			628	815				60%				770	589
			607	786				65%				718	548
			574	741				70%				651	495
			529	680				75%			780	568
				603	804			80%			647	468
				408	675			85%		725	492
					525	725		90%	770	474

	fewer than 650 cases needed, i.e., proposed study (650 cases and 650 controls) would be able to detect a difference
	more than 650 cases needed, i.e., proposed study would not have enough power to detect a difference

Alpha = 5%; Beta = 20% (power=80%)

Assuming 1:1 case to control ratio

For example: if 25% of the control (not injured) immediately leaves the residence, but only 18% of the case (injured) group does, we would need 568 individuals in both the case and control group to reliably detect a statistical difference. Our study sample size for each group (650) meets these criteria. However, if the difference between the groups is smaller (25% versus 17%), we would need greater than 780 individuals in each group. In that scenario, we would lack the precision and statistical power to provide a reliable answer about whether or not the behaviors in these two groups (cases and controls) are significantly different.