So for this analysis I have exluded masks with codes 3,6,7,11 here the labels . label define mask 1 "Alpha Solway 3030V" 2 "Alpha Solway HX-3" 3 "Alpha Solway S-3V > " 4 "CORPRO HM1400" 5 "Dräger 1730" 6 "Dräger 1730V" 7 "Fang Tian FT-045A" 8 "GVS F > 31000" 9 "Handanhy HY9330" 10 "Handanhy HY9632" 11 "Honeywell Super One" 12 "Medium > Kolmi" 13 "Meixin MX-2016V" 14 "Small Kolmi" 15 "Valmy Spireor VSP352TF", replace . table mask ----------------------------------- Question_3_Type_of_Mas | k | Freq. -----------------------+----------- Alpha Solway 3030V | 2,552 Alpha Solway HX-3 | 7,228 Alpha Solway S-3V | 315 Dräger 1730 | 7,076 Dräger 1730V | 195 Fang Tian FT-045A | 73 GVS F31000 | 10,355 Handanhy HY9330 | 6,481 Handanhy HY9632 | 4,134 Honeywell Super One | 166 Medium Kolmi | 2,632 Meixin MX-2016V | 2,893 Small Kolmi | 1,620 Valmy Spireor VSP352TF | 1,236 ----------------------------------- Evidently, this catches all the low numbers. Now here is the analyis for the effects of masks adjusting for gender, age and thnic group stratified by trust. . by trust, sort : logistic fit ib(8).mask ib(1).sex ib(1).ethnic ib(1).age if mask ! > = 3 & mask !=6 &mask !=7 & mask !=11, baselevels ------------------------------------------------------------------------------------- -> trust = East of England Logistic regression Number of obs = 3,994 LR chi2(18) = 110.83 Prob > chi2 = 0.0000 Log likelihood = -1608.2305 Pseudo R2 = 0.0333 ------------------------------------------------------------------------------------ fit | Odds Ratio Std. Err. z P>|z| [95% Conf. Interval] -------------------+---------------------------------------------------------------- mask | Alpha Solway 30.. | 5.575929 2.581968 3.71 0.000 2.2499 13.81883 Alpha Solway HX-3 | .9298395 .1622267 -0.42 0.677 .660543 1.308925 Dräger 1730 | 1.178419 .1759317 1.10 0.271 .8794681 1.578991 GVS F31000 | 1 (base) Handanhy HY9330 | .7671833 .1226266 -1.66 0.097 .5608453 1.049434 Handanhy HY9632 | 1.834741 .2777179 4.01 0.000 1.36374 2.468413 Medium Kolmi | .5919122 .0915989 -3.39 0.001 .4370521 .8016437 Meixin MX-2016V | 2.302081 .4738405 4.05 0.000 1.53786 3.446074 Small Kolmi | .8827132 .256725 -0.43 0.668 .4991825 1.560917 Valmy Spireor V.. | .4765458 .2365582 -1.49 0.135 .180122 1.26079 | ------------------------------------------------------------------------------------- -> trust = London Logistic regression Number of obs = 6,463 LR chi2(18) = 282.52 Prob > chi2 = 0.0000 Log likelihood = -2496.2356 Pseudo R2 = 0.0536 ------------------------------------------------------------------------------------ fit | Odds Ratio Std. Err. z P>|z| [95% Conf. Interval] -------------------+---------------------------------------------------------------- mask | Alpha Solway 30.. | 17.08056 7.79954 6.21 0.000 6.979365 41.80117 Alpha Solway HX-3 | .6828092 .0859175 -3.03 0.002 .5335721 .8737871 Dräger 1730 | .5479712 .0619715 -5.32 0.000 .439029 .6839468 GVS F31000 | 1 (base) Handanhy HY9330 | .9891796 .1293929 -0.08 0.934 .7654748 1.278261 Handanhy HY9632 | 1.044059 .1613928 0.28 0.780 .7711603 1.413532 Medium Kolmi | .4768866 .0657527 -5.37 0.000 .3639587 .6248533 Meixin MX-2016V | 2.188966 .523798 3.27 0.001 1.369478 3.498833 Small Kolmi | .6613201 .096202 -2.84 0.004 .4972647 .8794997 Valmy Spireor V.. | 1.385011 .3807315 1.18 0.236 .8080954 2.373797 ------------------------------------------------------------------------------------- -> trust = Midlands Logistic regression Number of obs = 11,956 LR chi2(18) = 407.82 Prob > chi2 = 0.0000 Log likelihood = -5781.3974 Pseudo R2 = 0.0341 ------------------------------------------------------------------------------------ fit | Odds Ratio Std. Err. z P>|z| [95% Conf. Interval] -------------------+---------------------------------------------------------------- mask | Alpha Solway 30.. | 5.01172 .7926168 10.19 0.000 3.675928 6.832924 Alpha Solway HX-3 | .9698577 .0671103 -0.44 0.658 .8468535 1.110728 Dräger 1730 | 1.596433 .147662 5.06 0.000 1.331738 1.913738 GVS F31000 | 1 (base) Handanhy HY9330 | .8340542 .067902 -2.23 0.026 .7110436 .9783456 Handanhy HY9632 | 1.001466 .0872105 0.02 0.987 .8443277 1.187849 Medium Kolmi | .6228992 .0649379 -4.54 0.000 .5077842 .7641109 Meixin MX-2016V | .7237203 .0803063 -2.91 0.004 .5822624 .8995448 Small Kolmi | .5495417 .0662517 -4.97 0.000 .4338918 .6960171 Valmy Spireor V.. | .3858165 .0761283 -4.83 0.000 .262074 .5679859 | ------------------------------------------------------------------------------------- -> trust = North East and Yorkshire Logistic regression Number of obs = 8,286 LR chi2(18) = 161.76 Prob > chi2 = 0.0000 Log likelihood = -3411.4356 Pseudo R2 = 0.0232 ------------------------------------------------------------------------------------ fit | Odds Ratio Std. Err. z P>|z| [95% Conf. Interval] -------------------+---------------------------------------------------------------- mask | Alpha Solway 30.. | .2004493 .103107 -3.12 0.002 .0731415 .5493451 ************** Alpha Solway HX-3 | .7695188 .1013947 -1.99 0.047 .5943766 .9962693 Dräger 1730 | .5531543 .0550147 -5.95 0.000 .455186 .672208 GVS F31000 | 1 (base) Handanhy HY9330 | .4953562 .0470393 -7.40 0.000 .4112322 .5966892 Handanhy HY9632 | .7218849 .1518303 -1.55 0.121 .4780125 1.090176 Medium Kolmi | .7418987 .1002188 -2.21 0.027 .5693256 .9667819 Meixin MX-2016V | .3827213 .0570472 -6.44 0.000 .2857626 .5125779 Small Kolmi | .9480932 .1813088 -0.28 0.780 .6517354 1.379211 Valmy Spireor V.. | .2527502 .0466354 -7.45 0.000 .1760495 .3628677 | ------------------------------------------------------------------------------------- -> trust = North West Logistic regression Number of obs = 3,741 LR chi2(18) = 110.44 Prob > chi2 = 0.0000 Log likelihood = -1399.1037 Pseudo R2 = 0.0380 ------------------------------------------------------------------------------------ fit | Odds Ratio Std. Err. z P>|z| [95% Conf. Interval] -------------------+---------------------------------------------------------------- mask | Alpha Solway 30.. | 4.403682 1.288689 5.07 0.000 2.481539 7.814672 Alpha Solway HX-3 | .8088528 .1145271 -1.50 0.134 .6128387 1.067561 Dräger 1730 | 1.132893 .1721026 0.82 0.411 .841161 1.525803 GVS F31000 | 1 (base) Handanhy HY9330 | .5379491 .109311 -3.05 0.002 .3612247 .8011336 Handanhy HY9632 | .4821311 .1435252 -2.45 0.014 .269012 .8640891 Medium Kolmi | .4984965 .1613906 -2.15 0.032 .2642905 .9402486 Meixin MX-2016V | .5042504 .1013884 -3.41 0.001 .3400144 .7478168 Small Kolmi | .526978 .3527976 -0.96 0.339 .1418849 1.957262 Valmy Spireor V.. | .5423083 .212063 -1.56 0.118 .2519964 1.167074 | ------------------------------------------------------------------------------------- -> trust = South East Logistic regression Number of obs = 11,048 LR chi2(18) = 909.69 Prob > chi2 = 0.0000 Log likelihood = -6155.2323 Pseudo R2 = 0.0688 ------------------------------------------------------------------------------------ fit | Odds Ratio Std. Err. z P>|z| [95% Conf. Interval] -------------------+---------------------------------------------------------------- mask | Alpha Solway 30.. | 4.381435 1.11839 5.79 0.000 2.656694 7.225887 Alpha Solway HX-3 | .7887919 .0688112 -2.72 0.007 .6648242 .9358754 Dräger 1730 | .3697371 .0306613 -12.00 0.000 .3142718 .4349916 GVS F31000 | 1 (base) Handanhy HY9330 | .2675777 .0211894 -16.65 0.000 .2291097 .3125045 Handanhy HY9632 | .2453184 .0203347 -16.95 0.000 .2085321 .2885939 Medium Kolmi | .8142324 .1055556 -1.59 0.113 .631539 1.049776 Meixin MX-2016V | .2049324 .018591 -17.47 0.000 .1715504 .2448103 Small Kolmi | 1.101414 .1748755 0.61 0.543 .8068676 1.503485 Valmy Spireor V.. | .4108836 .0415114 -8.80 0.000 .3370716 .5008589 | So, the most important message is that the best performing mask is Alpha Solway 3030V. This is consistent over all trusts. Exception is North East and Yorkshsire, but here this type was NOT used, see below. So, I would take that one out for that trust. Probably the next best is GVS F31000. ------------------------------------------------------------------------------------ -> trust = North East and Yorkshire Question_3_Type_of_Mas | k | Freq. Percent Cum. -----------------------+----------------------------------- Alpha Solway 3030V | 17 0.20 0.20 Alpha Solway HX-3 | 754 9.05 9.26 Dräger 1730 | 1,727 20.74 30.00 Dräger 1730V | 12 0.14 30.14 Fang Tian FT-045A | 2 0.02 30.16 GVS F31000 | 2,045 24.56 54.72 Handanhy HY9330 | 1,993 23.93 78.65 Handanhy HY9632 | 231 2.77 81.42 Honeywell Super One | 3 0.04 81.46 Medium Kolmi | 698 8.38 89.84 Meixin MX-2016V | 354 4.25 94.09 Small Kolmi | 328 3.94 98.03 Valmy Spireor VSP352TF | 164 1.97 100.00 -----------------------+----------------------------------- I also had no results left for the South West trust due to the limitations.