lasso retained features with new coding - `presence`: average strategy presence/use ```r feature b presence <char> <num> <num> 1: audience_adaptation 0.614 0.190 2: contrastive_tone2 0.469 0.300 # new coding 3: evidence_or_factbased_arguments 0.360 0.642 4: conditionZ:preZ 0.276 NA 5: transfer_of_association 0.256 0.228 6: topicZ 0.238 NA 7: politeness_and_civil_tone 0.233 0.974 8: negative_tone2 0.187 0.072 # new coding 9: positive_tone2 0.171 0.321 # new coding 10: namecalling 0.077 0.095 11: relatable_hypotheticals 0.069 0.092 12: encourage_action_with_clear_calls 0.066 0.250 13: topicZ:preZ -0.016 NA 14: conditionZ:topicZ -0.106 NA 15: emotional_appeal_with_balanced_urgency -0.142 0.223 16: stimulate_anger -0.300 0.055 17: address_objections_and_counterarguments -1.014 0.451 18: preZ -2.183 NA # without z-scoring all the strategies feature b presence <char> <num> <num> 1: audience_adaptation 2.393 0.190 2: contrastive_tone2 1.658 0.300 3: evidence_or_factbased_arguments 1.229 0.642 4: transfer_of_association 1.041 0.228 5: conditionZ:preZ 0.434 NA 6: positive_tone2 0.300 0.321 7: topicZ 0.289 NA 8: encourage_action_with_clear_calls 0.282 0.250 9: conditionZ -0.002 NA 10: conditionZ:topicZ:preZ -0.020 NA 11: topicZ:preZ -0.147 NA 12: conditionZ:topicZ -0.182 NA 13: preZ -2.260 NA 14: address_objections_and_counterarguments -3.359 0.451 ``` for comparison, lasso retained features with old coding: - features included in above but not below: - `topicZ:preZ`, `namecalling` ```r feature b presence <char> <num> <num> 1: audience_adaptation 0.581 0.190 2: contrastive_tone 0.378 0.610 # original LLM coding (variable was excluded in original lasso) 3: evidence_or_factbased_arguments 0.292 0.642 4: topicZ 0.258 NA 5: negative_tone 0.234 0.372 # original LLM coding 6: positive_tone 0.220 0.622 # original LLM coding 7: conditionZ:preZ 0.204 NA 8: politeness_and_civil_tone 0.195 0.974 9: transfer_of_association 0.169 0.228 10: relatable_hypotheticals 0.053 0.092 11: encourage_action_with_clear_calls 0.016 0.250 12: conditionZ:topicZ -0.071 NA 13: emotional_appeal_with_balanced_urgency -0.128 0.223 14: stimulate_anger -0.262 0.055 15: address_objections_and_counterarguments -0.948 0.451 16: preZ -2.175 NA # without z-scoring all the strategies feature b presence <char> <num> <num> 1: audience_adaptation 2.342 0.190 2: contrastive_tone 1.493 0.610 3: positive_tone 1.221 0.622 4: evidence_or_factbased_arguments 0.938 0.642 5: negative_tone 0.771 0.372 6: transfer_of_association 0.663 0.228 7: topicZ 0.413 NA 8: conditionZ:preZ 0.335 NA 9: encourage_action_with_clear_calls 0.304 0.250 10: conditionZ 0.058 NA 11: topicZ:preZ -0.149 NA 12: conditionZ:topicZ -0.204 NA 13: emotional_appeal_with_balanced_urgency -0.272 0.223 14: preZ -2.525 NA 15: address_objections_and_counterarguments -3.263 0.451 ```