Table 2 Variables and description
From: Offline visit intention of online patients: the Grice’s maxims and patient involvement
Variable type | Variable name | Measurement | Description | Abbreviation |
|---|---|---|---|---|
Dependent variable | Offline visit intention | The number of sentences containing the offline visit intention | We manually labeled the sentences of patients’ posts. Then, we used the labeled data and machine learning method to get the classifier. And used the classifier to predict the patient’s offline visit intention from sentences. | Intention |
Independent variables | Amount of information. | The logarithm of the quantity of content words. | Content words: the adjectives, nouns, numerals, quantifiers, pronouns, and verbs. [70] | Content |
The logarithm of the quantity of unique words | The unique words in the physician’s posts of the thread (Robust test) | Unique | ||
Reliability | Objective sentences ratio | Ratio of the number of objective sentences to the total number of sentences in replies. [68] [74] | Reliability | |
Relevance | Topics relevance between the posts of the patient and the physician | LDA (an unsupervised machine learning algorithm, Appendix D) classified and predicted the topics and probabilities of physician-patient posts, respectively, and calculated the topics’ relevance. | Relevance | |
Understandability | Average sentence segments length. | The average sentence segment length is calculated by dividing the number of words by the number of semicolons, commas, periods, question marks and exclamation marks, | Length | |
Volume of positive e-WOM | The heat index of the physician | The heat is calculated based on the number of patients recommended in the past two years and converted into a decimal value of 1 to 5 | Heat | |
Thank-you letters | Number of thank-you letters received by physicians (Robust test) | letters | ||
Expertise cue | The logarithm of clinic titles of physician | Chief physician 2,associate physician 1,other 0 | Title | |
Hospital reputation | Whether the hospital is a top-tier Grade A | If the hospital is top-tier Grade A then 1, else 0. | Hospital | |
Moderator variable | Involvement | The number of patient’s topics | Calculate the number of patient topics after predicting the patient topics and probabilities by LDA (Appendix D), | Topics |
Control variable | Disease risk | Whether the disease risk is high, dummy variable. | According to the mortality rate of the disease , the diseases are classified into high-risk disease or low-risk disease categories, and all malignant tumors are classified as high risk. | Risk |
Response rate | Response rate | The ratio of the number of physician’s posts to that of patient in a thread. | Response | |
Treat experience | Treat experience | Extract from the patient’s posts whether the patient has visited the other physician offline before. | Before | |
Duration of illness | The logarithm of duration of illness | The duration was extract from the posts, 1 means less than a week, 2 means less than a month, 3 means less than six months, 4 means more than six months | Duration | |
Tel service | The telephone service | Whether the physician provides telephone service. If provided,1, else 0. | Tel | |
Transfer services | The transfer service | Whether the physician provides the transfer service. If provided,1, else 0. | Transfer | |
Page view | Page view | The physician’s page’s view times | View | |
Price | The logarithm of web consulting price. | Patients need to pay a fee to use the physician’s online counseling service | Price | |
City scale | The logarithm of the city scale | The scale of the city where the physician works, based on “2018 China city business charm list” | City |